In the forest in Mandena, Madagascar, banana-leafed Ravenala trees crowd out the sun, their electric blue seed pods dotting the leaf litter and white sand below. When night falls, gray mouse lemurs emerge from tree hollow dens to feed on insects, flowers and fruit. During the rainy season, pools of water form where screw pines’ pom-pom-like clusters of long leaves meet their trunks, the base of each leaf forming a reservoir just large enough to nurture small schools of tadpoles to maturity before the puddles dry out every April. There ring-wearing tree frogs—named for the bright-white bands that mark each webby finger—find a perfect spot to nurture their next generation, high above would-be predators. Leopard-spotted and no bigger than a child’s thumb, the frogs lay their eggs in a sticky clutch above the water and stand watch for nearly a week, until their offspring drop into the tiny pool and begin to swim.

At close range, this corner of Mandena feels like you could get lost in it. But above the canopy reality looms into view. Forest once stretched to the horizon. What’s left of it is now smaller than Brooklyn’s Prospect Park—less than a half-hour walk from end to end, sandwiched between a mine on one side and a steadily expanding village on the other.

Madagascar broke free of the land that makes up Africa and India nearly 100 million years ago. Across the eons, evolution in isolation has given the island unparalleled ecological richness: Four out of five plants and animals there are found nowhere else, the sweeping cast of characters in a wide array of highly specialized symbiotic niches. The country’s 83 species of screw pine alone serve as breeding grounds for dozens of different reptiles and amphibians. But the ballet between this particular tree and frog is now confined to a tiny collection of forest fragments, like the one in Mandena, that are spread along Madagascar’s southeastern coast. Two of the three smatterings of forest where the frog is still found lie inside a concession belonging to Rio Tinto, one of the largest mining companies in the world.

Rio Tinto came to Madagascar in the 1980s, looking for ilmenite, a mineral used to make titanium dioxide, which provides the white pigment found in products ranging from paint and plastics to toothpaste. Test pits hit pay dirt near Tolagnaro (Fort Dauphin), at the southeastern tip of the island. The ilmenite deposits that interest the company lie underneath the remnants of dense evergreen forests that once grew on sand dunes along most of Madagascar’s eastern coast, forming a continuous band covering perhaps 465,000 hectares. Since human colonization of the island some 2,000 years ago, these littoral forests, as they are known, have dwindled to at most 10 percent of their original expanse. As such, Rio Tinto’s concession weaves through one of the most threatened ecosystems on the planet.

Ordinarily, the discovery of so much buried wealth underneath an already vulnerable ecosystem would spell doom for most of what lives there. But in 2004 executives at Rio Tinto, which is headquartered in London, flew to the International Union for Conservation of Nature’s World Conservation Congress in Bangkok, a major gathering of scientists, environmentalists, and government and business leaders, to unveil what amounted to a radical rethinking of mining’s relationship with the natural world. Going forward, they pledged, the company would seek not just to limit the environmental damage it caused but to actively improve the ecology of its most sensitive mine sites. And it would start with the mining concession in southeastern Madagascar.

Sun rises over Tolagnaro (Fort Dauphin), Madagascar. Some 70 million metric tons of ilmenite lie under the littoral forest in this region. Credit: Martelet Christian Getty Images

Conservationists met the proposal with enthusiasm. They had reason to be optimistic: Rio Tinto and its predecessor had already been collaborating with scientists from the Missouri Botanical Garden for more than a decade, funding and conducting botanical surveys and studies of the new species discovered throughout the company’s concession. There were few details yet and no hard benchmarks, but if Rio Tinto followed through, the stance had the potential to reverberate throughout the industry, forcing mining companies to compete for permits on the basis of their environmental programs.

As part of this conservation initiative, Rio Tinto had created what the company called a biodiversity committee made up of researchers and nonprofit managers who could help its local subsidiary, QIT Madagascar Minerals (QMM), plan and carry out environmental work on the margins of what promised to be an enormous mine. Madagascar’s government would receive a 20 percent stake in QMM—an investment that could generate hundreds of millions of dollars in new revenue for the country over time. For the scientists in the group, joining the committee represented a leap of faith. Their input could prevent the worst and harness Rio Tinto’s investment for environmental good. But it also meant they would share the blame for anything that went wrong.

It didn’t take long. Within a few years of the committee’s inception, its members repeatedly raised concerns that QMM was not on track to meet its biodiversity goals. When ilmenite prices slumped during the Great Recession, Rio Tinto’s priorities shifted, and by 2016, the company reneged on its grand conservation promise. Instead it adopted the vague goal of avoiding making things too much worse. Today mining near Mandena is poised to extinguish this biodiversity hotspot. For the people who live there and dozens of endemic species such as the ring-wearing tree frog, destiny now turns on the outcome of this long-running experiment, a test case for industry’s role in conservation and the role conservationists can play in the mining industry.

In its natural state, ilmenite accumulates in the deep sediments deposited by rivers and streams that changed course long ago, forming a black sand so heavy it separates from lighter minerals at the surface. To extract the mineral, miners begin by using backhoes and chain saws to remove every scrap of vegetation from each mining tract and pile it into gargantuan mounds of compost. Earth-moving machines dig a trench several stories deep and longer than a football field, which is then filled with water diverted from a nearby river. A dredge stirs up sand from a depth of up to 18 meters and pumps it onto a barge through an oversized straw, where gravity separates some of the ilmenite ore from sand, topsoil and lighter materials. Great “black snakes”—temporary pipelines—crisscross the expanse, conveying the mineral-rich slurry to a gleaming green processing plant near the water. Electrostatic separation is used to extract still more ilmenite before the demineralized sand and soil are spread back out over the landscape.

Rio Tinto discovered ilmenite near Tolagnaro in 1986. At the time, the forests in the region were already heavily fragmented and degraded by human activity. But the company’s prospecting soon brought new roads to the area and an influx of people looking for work, hastening the deforestation underway for charcoal production and new farmland to supply the growing city.

Rio Tinto's mining sites in Madagascar are home to a number of imperiled species, including the ring-wearing tree frog (left, top), the Antanosy day gecko (leftbottom) and the collared brown lemur (right). Some species are known only from areas within the company's mining concession. Credit: Franco Andreone (frog and gecko); Konrad Wothe Nature Picture Library (lemur)

Rio Tinto determined that the region around Tolagnaro contained some 70 million metric tons of ilmenite—enough to supply about 10 percent of the global market for a decade or more—and began to make a plan for extracting it. The company set its sights on three mineral-rich areas along the coast encompassing a total of approximately 6,000 hectares. Mining would start at the 2,000-hectare site in Mandena and eventually expand north to Sainte Luce and to Petriky farther south. The extraction would continue for the life of the mine—about 60 years from the date of first production, according to the company’s projections. Rio Tinto estimated that in the end the project would result in the loss of 1,665 hectares, or 3.5 percent, of Madagascar’s remaining littoral forest.

While Rio Tinto explored the area to gauge the full extent of the ilmenite deposits, it initiated environmental studies. As part of this effort, the company funded one of the first botanical inventories of forests along Madagascar’s eastern coast—Rio Tinto knew it would stand a better chance of securing the requisite mining permits if it could show that it had done due diligence about the extent of environmental damage its mining would bring about. Botanist Pete Lowry worked with a team of his colleagues at the Missouri Botanical Garden to collect and document every plant species they encountered. As the team found dozens of unfamiliar plants, Lowry says, “it sort of dawned on us—there are a lot of species that seem to grow on white sand and nowhere else.” The team was tracing the outlines of an ecosystem scarcely known to science. Rio Tinto went on to partner with top-flight researchers from around the world, supporting studies on more than 40 previously undescribed species found in the mining concession.

Despite Rio Tinto’s support for ecological research in Madagascar, by the early 2000s the company’s global track record had earned it a reputation as an unscrupulous actor in a heavily polluting industry. In Papua New Guinea, where Rio Tinto had developed a giant copper mine in the 1980s, protests brought on by the company’s disparate treatment of white foreigners and local workers forced the mine’s closure and helped to spark a civil war. Thirty years later Rio Tinto is gone, but pollution from the shuttered Panguna mine will still cost an estimated $1 billion to clean up.

It was against this troubled backdrop that Rio Tinto went to Bangkok in 2004 to announce a pilot conservation initiative in Madagascar. The company called the strategy net positive impact (NPI). It pledged to leave the local ecosystems in Mandena, Sainte Luce and Petriky—all of which have especially high biodiversity—better off because of mining than they would have been without it. In 2005 Rio Tinto began to roll out the particulars of its plan. It would avoid mining altogether in well-preserved forest fragments in each of the three sites; undertake unprecedented ecological restoration of areas cleared during mining; and invest in biodiversity offsets at several forest sites elsewhere in the region to compensate for the damage it would do in the mining zone. The biodiversity committee would serve to help the company make good on its promise.

Women harvest mahampy, a type of reed that grows in wetlands along the coast. The mahampy is coated with clay and dried before it is woven. For villagers near Mandena, one of Rio Tinto's mine sites, woven mahampy has long been a key source of income. But the supply of the reed has dwindled as wetlands there fill with sand from mining. Credit: Ed Kashi Redux Pictures

The partnership did not sit well with some conservationists. Barry Ferguson, an environmental researcher then based in Tolagnaro, saw the arrangement as a kind of mutually beneficial “greenwashing” whereby scholars with strong conservation bona fides boosted their research careers with studies funded by QMM. Other observers were skeptical that net positive impact was a target Rio Tinto could ever meet in such an ecologically sensitive area. After all, dozens of plant species are known only from areas within the mining concession. The existence of a particular species of day gecko, Phelsuma antanosy, a tiny dart of neon green with red stripes and flashes of turquoise on the males, is even more precarious. Confined to habitat thought to be less than 10 square kilometers, the gecko lays its eggs on a single species of screw pine and forages for insects on the same tree.

Achieving NPI in Madagascar would be an expensive proposition. Rio Tinto calculated that it would have to leave $1.2 billion of ilmenite underground to spare the 624 hectares of forest in the so-called avoidance zones and convert them into protected areas. Restoring ruined forest and creating offsets would cut further into its profits.

Yet in promotional materials, the company often argued the “business case” for NPI, based on a need to show governments and investors that Rio Tinto is the best firm to carry out projects with major social and environmental risks. The way Lowry understood it, “occupying the high ground would give [Rio Tinto] a commercial advantage.” He became the biodiversity committee’s president in 2006. Early on Lowry hoped the mine in Madagascar, along with two other Rio Tinto pilot sites for NPI in Mongolia and Australia, could help define a new path for the mining industry’s relationship with the environment at a time when companies were concerned that social and environmental risks might lock them out of potentially lucrative sites. “The idea was, We’re a dirty business, everybody knows we’re a dirty business,’” he says. ‘“What do we need to do to gain access in the future?’” Rio Tinto officially began mining operations there in 2008.

But the business case for NPI soon ran headlong into the business of running a profitable mine. Global financial markets plunged in the months before Mandena was set to enter production in December 2008, and Rio Tinto’s stock price tumbled as the company braced for lower demand. The first shipments of ilmenite left Madagascar for processing in Canada in May 2009; by the end of the year demand for the mineral was down 20 percent.

For a while Rio Tinto upheld part of its conservation promise, steering clear of its designated avoidance zones. But simply avoiding these protected areas was not enough—the forests were continuing to degrade from lack of active management and encroaching loggers and charcoal producers. The biodiversity committee grew concerned that the company was not ramping up its conservation work accordingly. “Species extinction is QMM’s biggest biodiversity risk,” the committee warned in 2010.

The conservation outlook deteriorated from there. Between 2010 and 2012 QMM was supposed to have made substantial progress in adding to forest cover through restoration work. Instead data from the company’s own incremental reviews show that deforestation had already claimed an area nearly as large as the protected one in Mandena. One important fragment, in Sainte Luce—home to four of the seven critically endangered species present in QMM’s mining footprint—was on pace to dwindle from more than 200 hectares to less than 50 hectares by 2024. The warnings captured in the minutes of the biodiversity committee’s meetings grew more urgent: “HUGE RISK FOR ACHIEVING NPI,” members wrote in 2012, arguing that QMM was running out of ways to offset future damage done by the mine.

Meanwhile a series of technical snafus in Madagascar and a costly investment blunder in Mozambique, where Rio Tinto overpaid for a stake in a massive new coal mine, ate into the company’s bottom line, prompting cost-cutting measures across the enterprise. Although the environmental program’s funding was not facing cuts, it seemed to be falling behind any realistic shot at NPI. Months were lost as Rio Tinto pushed for QMM to shoulder more responsibility for funding the work on its own budget.

Even as the mining dredge steadily ate away at the other fragments in Mandena, QMM had successfully curbed deforestation in the protected area to near zero. But Mandena is by far the easiest of the three sites to manage and the least important for biodiversity. By 2015 QMM’s Biodiversity Action Plan warned that achieving NPI required immediately stopping degradation and deforestation in both the offset and avoidance zones in Petriky and Sainte Luce and dramatically slowing the loss of forest in the offsets outside the mining area.

Then, in 2016, Rio Tinto officially abandoned NPI as a corporate mandate. A representative met with QMM’s biodiversity committee to present a new corporate environmental standard set to replace NPI, one it framed as “minimizing residual impact.” What, exactly, did that mean?

“It was totally devoid of anything really substantive,” Lowry recalls. The most Rio Tinto will say publicly is that the answer is “site-specific”: individual projects can define and pay for their own environmental management—up to and, if they wish, including net positive impact.

Jörg Ganzhorn, an ecologist at the University of Hamburg in Germany who had been collaborating with Rio Tinto and QMM for more than a decade, was stunned. “I would understand if you as a mining company do not claim net positive biodiversity impact. That’s not your job,” he says. But no one had forced Rio Tinto to tout the standard on its Web site and fly its CEO to environmental conferences around the globe to speak about the company’s groundbreaking initiative. To do all that and then abandon NPI? “That’s when I decided I had to leave,” Ganzhorn says. That October, he, Lowry and the two other remaining scientists advising Rio Tinto in Madagascar released a statement abruptly cutting ties with the company.

Soon afterward, Rio Tinto executives circulated a set of talking points responding to the committee’s resignation, portraying its undoing as a mutual agreement “to refresh the objectives and focus of the panel.” A new and improved committee would be formed, with former members lending a hand to shape its work, the statement said. Lowry was the only former member still open to being involved going forward. “The stakes are still very high,” he says. “If I don’t serve on this committee, there will be zero connection to the work that’s been done over the past 20 to 25 years.”

In July 2017 I rode along with two members of QMM’s environmental team on a tour of Mandena, where a patchwork of rolling fields, forest fragments and wetlands is steadily giving way to the hard corners and straight lines of an industrial site. A Madagascar kestrel perched on a fence post. Rows of eucalyptus and acacia saplings formed a grid over the sandy expanse where the mining dredge had passed. Over time QMM hopes these trees will provide a source of wood and charcoal for communities that currently depend on forest fragments that will soon be mined. Just behind the company’s headquarters, QMM maintains a nursery that supplies it with acacia and eucalyptus, along with native plants it is using in experiments aimed at restoring some 675 hectares of forest by the end of the mine’s life in 2065.

A family of eastern lesser bamboo lemurs (Hapalemur griseus) frolicked by an outbuilding, gnawing on bamboo shoots, as Faly Randriatafika, who oversees QMM’s environmental work, walked through rows of tiny seedlings arranged in plastic trays. He pointed to an eight-centimeter sapling of Eligmocarpus cynometroides, a spindly palm with fist-shaped seeds, represented by about 20 specimens in the wild, all confined to Petriky. “This plant is very hard to germinate: out of 500 fruits, you get maybe only 20 seeds,” he said. “Without QMM, without this project, this species would have disappeared completely.”

Lisa Gaylord, then the company’s manager of corporate relations, communities and sustainable development, made a similar observation about the fate of the littoral forests around QMM’s mine more broadly. At QMM’s satellite office in Tolagnaro, she pulled out her laptop to show me an animated slide depicting changes in forest cover around Sainte Luce over the preceding decade. The patches of green shrank from year to year like sandbars disappearing below a high tide. The implication was clear: mine or no mine, charcoal making and farming will soon take over what little forest remains. “We could do nothing, and I could tell you, that entire forest corridor will go,” she said. “It will go. That’s where Madagascar’s going.”

Yet there can be no doubt that the mining is taking a grave toll—not only on forests and wildlife but on people. A village lies at the top of a small hill above the mining area in Mandena, along a rutted dirt road known as the old highway, less than half a kilometer inland from the smooth tarmac road QMM built for its own private use. The cheffokontany, or local “headman,” Francis Maka Teodorik, gathered 10 of his neighbors to talk with me in his home, where we sat on traditional mats made from mahampy, a type of reed gathered in wetlands up and down the coast. Woven mahampy has long been the dominant source of income for women here, and along with timber for construction, fuel and charcoal making, its supply is shrinking.

QMM has funded a demonstration plot of restored wetlands and training sessions to encourage local women to harvest mahampy sustainably by cutting above the roots. But Teodorik and his neighbors said these efforts obscure the real impact of QMM’s mine. Helenette Raverosaotra, a mother of four whose two-room house overlooks QMM’s processing plant, said it now takes as many as six or seven foraging trips, instead of one, to collect enough reeds to weave a mat that sells for less than $3, as the wetlands around Mandena have been mined one by one. “QMM has already destroyed all the mahampy we used for mats,” said Fidéline Jine, who now spends her days fishing for shrimp in the river to earn a small fraction of what she once made. “The mines have filled with sand all the places where the mahampy grew.”

Local farmers, whose land was flooded to create a water source for the mine, had another grievance. For years they protested that they had not received fair compensation for the amount of land they lost. When QMM finally agreed to assess how much farmland it had taken over, the company’s own analysis showed the farmers were right—QMM had paid the farmers for the loss of four hectares but had taken more than six times that amount. QMM eventually paid the farmers for the balance.

One missing ingredient from the mining-conservation partnership, everyone seems to agree, is more robust government oversight. Says Jocelyn Rakotomalala, who runs a Tolagnaro-based NGO called Saha, which works with QMM on social and community projects in the area: “Mining companies could conserve more if only the state were more demanding.”

Rio Tinto has often credited its commitment to NPI as a crucial factor in gaining approval for the project, but as Heritiana Ravelojaona, the provincial director of mining in the region, points out, the agreement it signed with the Malagasy government does not require anything like NPI. “Take the case of the offsets,” he says. “Those are voluntary commitments.” And in Sainte Luce, where villagers have repeatedly protested their loss of access to the small protected areas created by the project, he says, “it’s no longer QMM’s business. It’s up to the state, if it decides to protect the area, to come up with a way to help satisfy the demands of the community after restricting access.”

Frank Hawkins, who now runs the Washington, D.C., office of the International Union for Conservation of Nature, was one of the first scientists to become involved with QMM. He now feels that QMM has been a “dismal failure” in terms of social and environmental outcomes. But Hawkins says he would still get involved if the process started over again today because the probable alternative to Rio Tinto is not no mines but mines built with woefully inadequate environmental protections. The planet is already littered with examples. In Butte, Mont., in 2016 thousands of snow geese were killed when a storm drove them into a toxic reservoir left behind by an open-pit copper mine that had ceased operations decades earlier. In the Niger River delta, oil exploration has brought the equivalent of an Exxon Valdez spill every single year for 50 years. “The sad truth is that the mining sector finds it very easy to negotiate big deals because you’re always talking about lots of money,” Hawkins explains.

In Ampasindava, a peninsula in northwestern Madagascar, high-level Malagasy officials have appeared eager to grant approval to a rare-earth mining venture under investigation for financial misdealings, after it successfully lobbied to shrink an adjacent protected area. In southwestern Madagascar, an Australian firm is in the beginning stages of developing another large ilmenite mine, one that is likely to exacerbate water shortages in an arid ecosystem already straining under the pressures of drought and deforestation.

Few believe that the Malagasy government has the political will to extract more meaningful concessions from interested mine operators on the front end. Hawkins says he would like to see mining contracts negotiated in the context of broader regional development plans, so that tourism operators or conservation organizations might provide a counterweight and advocate for a broader vision of development.

For his part, Lowry is dismayed that Rio Tinto’s gambit on net positive impact does not seem to have spurred a new wave of competition among mining companies on environmental management. Indeed, the most hopeful signs of boosting industry’s environmental record in Africa have come the old-fashioned way, through government action. Chad, Sudan, Niger and Gabon, for instance, have all recently taken punitive action against SINOPEC and China National Petroleum Corporation, two state-owned Chinese oil giants, for pollution and exploitative management practices. Zambia got tough with coal-mining operations largely in response to local protests over labor conditions and pollution. Shortly after my visit in 2017, Malagasy officials made a fact-finding trip to a remote part of Rio Tinto’s concession to investigate community protests against the company—far more of a government reaction than the biodiversity committee got with its resignation letter.

Whether that reaction leads to any meaningful enforcement is a different question. Rio Tinto has acknowledged that mining in Mandena had encroached on a “buffer zone” around a lake that provides both mahampy and drinking water for communities nearby, increasing the risk that radioactive tailings left over from ilmenite extraction could seep into the water supply. The admission came only after two years of prodding by a British charity that works in the area, the Andrew Lees Trust, which had to commission a study by an independent geophysicist to prove the point. But it turns out that Madagascar’s environmental regulator—the National Office for the Environment, funded with fees from mining permits such as QMM’s—had known about the breach for at least a year. The office decided not to take any regulatory action.

The most reliable commitments from large mining projects seem to be those that come with money attached: In Mongolia, where the International Finance Corporation (IFC) owns a slice of the Rio Tinto project, net positive impact is still on the table—largely because it is attached to the IFC’s own performance standard on environmental stewardship. Elsewhere in Madagascar, some of the most successful environmental partnerships between the private sector and local communities are in the seafood industry, where there is a clearer link between end consumers in Europe and the ecological stakes of their purchasing decisions.

Still, Lowry does not regret his decision to work with Rio Tinto, even after seeing NPI collapse as a company-wide model. “I think where QMM is today is a whole lot better than where it would have been, in terms of environmental and social responsibility, if there had never been a committee,” he says. In 2018 Lowry chose to join QMM’s newly minted biodiversity and natural resources management committee to try and preserve some continuity with the previous group’s work. In a way, he was persuaded by Rio Tinto’s retreat. With QMM, at least, decisions about conservation won’t be made in London. From its offices in Tolagnaro, the forests are not an abstraction.